Petrogenesis of the Kabanga–Musongati layered
mafic–ultramafic intrusions in Burundi (Kibaran Belt):
geochemical, Sr–Nd isotopic constraints and Cr–Ni behaviour
Jean-Clair Duchesne
a,*, Jean-Paul Lie´geois
b, Andre´ Deblond
a,b, Luc Tack
b aDepartment of Geology, University of Lie`ge, B20, B-4000 Sart Tilman, BelgiumbRoyal Museum for Central Africa, B-3080 Tervuren, Belgium Available online 16 September 2004
Abstract
A succession of mafic–ultramafic layered intrusions forms an alignment in the boundary zone between the Kibaran belt and the Tanzania craton. The intrusions represent a continuous series of cumulate rocks. For instance, in the Mukanda-Buhoro and Muson-gati (MBM) contiguous bodies, the series starts with dunite and passes to lherzolite, pyroxenite, norite, gabbronorite and anortho-site on top. Cumulate textures are conspicuous in all rock types and cryptic layering characterises cumulus mineral compositions, thus evidencing fractional crystallization as a major differentiation mechanism. The increase of Cr in the ultramafic members of the series indicates that chromite was not a liquidus mineral in dunite and lherzolite rocks, thus unable to form chromitite layers. The high Ni-content of dunite seems to preclude the existence of conjugate Ni-rich sulphide deposits. The87Sr/86Sr initial ratio is rela-tively constant and averages 0.7087, with some values up to 0.712 due to local assimilation. Fine-grained rocks from the MBM area are isotopically (Nd and Sr) similar to the MBM cumulates. Modelling their crystallization produces cumulus mineral compositions similar to those in the Musongati ultramafic rocks, which suggests a broadly picritic parental magma. On the other hand, fine-grained rocks from the Nyabikere area are not related to the Nyabikere cumulates. Nd and Sr isotope ratios show that the MBM magmatism is related to an enriched source, possibly an old subcontinental lithospheric mantle. The Nyabikere dykes, as well as the Waga dykes, come from a depleted mantle source, as do the A-type granitoids occurring in the same boundary zone. Several lines of evidence point to two types of parental magmas, a picritic magma, and a more evolved magma, broadly similar to the Bush-veld Main Zone magma.
Ó 2004 Elsevier Ltd. All rights reserved.
Keywords: Burundi; Layered intrusions; Chromitite; Ni-sulphide; Chromite; Sr–Nd isotopes
1. Introduction
Layered gabbroic intrusions host the major Cr, Ni and PGE (platinum group elements) deposits (Naldrett et al., 1987; Cawthorn, 1996; Lee, 1996). It is now widely recognized (e.g. Evans, 1993) that the genesis of these orthomagmatic deposits is strongly controlled by the magma composition and magma chamber processes
such as fractional crystallization (with or without con-tamination), immiscibility, and magma mixing. Petro-logical studies aiming at defining these processes are thus useful to evaluate mining potential and to plan exploration methods.
The purpose of the present paper is to investigate the petrology and geochemistry (major and trace elements, Sr and Nd isotopes) of a series of Kibaran layered intru-sions in Burundi (Central Africa) in order to define mag-matic processes, which have controlled their genesis, as well as the nature of their parental magmas. Geochemi-cal characteristics of the rock series are also used to
0899-5362/$ - see front matter Ó 2004 Elsevier Ltd. All rights reserved. doi:10.1016/j.jafrearsci.2004.07.055
* Corresponding author. Tel.: +32 43 66 22 55; fax: +32 43 66 2921. E-mail address:jc.duchesne@ulg.ac.be(J.-C. Duchesne).
www.elsevier.com/locate/jafrearsci Journal of African Earth Sciences 39 (2004) 133–145
assess the metallogenic potential of the Burundian intru-sions. This approach complements the general evalua-tion of mineral resources presented by Deblond and Tack (1999).
2. Regional geological setting
The Northeastern Kibaran Belt in Burundi (Fig. 1) comprises a Western Internal Domain (WID), made up of a deformed sequence of Mesoproterozoic granites and amphibolite-greenschist facies metasedimentary and metavolcanic rocks, overthrust on an Eastern External Domain (EED). The latter comprises supracrustal mate-rial, which is less deformed and metamorphosed and
without granites, resting upon the Archaean Tanzania craton (Fig. 1; see references inTack et al., 1994, 2002). A series of mafic and ultramafic intrusions straddles the border (Fig. 1: ‘‘Boundary Zone’’) between the WID and EED domains, together with some A-type granitoids (Tack et al., 1994). These intrusions form the Kabanga–Musongati (KM) alignment, named after the Kabanga intrusion in Tanzania, explored for Ni-sul-phides (Evans et al., 1999, 2000), and the Musongati ultramafic intrusion in Burundi, a potential Ni lateritic deposit (Deblond and Tack, 1999and references therein;
Fig. 1). The emplacement age of the KM alignment is currently under debate. A previous 1275 Ma bulk zircon age (Tack et al., 1994) is challenged by a zircon
SHRIMP set of measurements spreading from
1370 Ma to 1200 Ma; the 1370 Ma end-member is inter-preted as the intrusion age (Wingate, zircon SHRIMP age in Tack et al., 2002). Preserved from pervasive deformation and metamorphism, the intrusions were only affected by brittle deformation associated with lo-cally developed thrust faulting. A contact aureole of cor-dierite- and andalusite-bearing spotted metapelites, overprinting the regional schistosity, attests to their intrusive character at relatively low pressure and pre-cludes an origin as part of a dismembered ophiolitic complex (Tack and Deblond, 1990).
Eight intrusions have been documented in Burundi (Deblond, 1993, 1994). They include (Fig. 1), from north to south, the Muremera, Nyabikere, Waga, Mukanda-Buhoro, Musongati, Rutovu, Nyange-Songa and Mugi-na intrusions. Among them, two broad types of intrusions can be defined: those which mainly consist of ultramafic rocks, such as Waga, Musongati, Nyabik-ere and Muremera, and those dominated by gabbroic
rocks, such as Mukanda-Buhoro, Nyange-Songa,
Rutovu and Mugina (Fig. 2, inset). In Tanzania, an updated distribution of the KM intrusions is given by
Evans et al. (2000).
3. The Mukanda-Buhoro and Musongati intrusions As an example, the Mukanda-Buhoro and Musongati contiguous bodies are focussed on because they associ-ate all rock types and were extensively studied (Deblond, 1994;Fig. 2). At Buhoro, the mafic rocks occupy depres-sions in the landscape and are often ferralitised. Out-crops are scarce, but the rocks, frequently occurring as boulders, are surprisingly fresh. In the eastern part of Musongati, a thick laterite covers the ultramafic rocks and sampling is only possible through drilled cores. Both bodies have been sliced by imbricate thrust faults. The Musongati body displays a series of thick layers, made up of different lithologies and dipping 60° on aver-age to the west. The lower part of the structure consists of ultramafic rocks (dunite and lherzolite grading
Fig. 1. Geological outline of the Northeastern Kibaran Belt, modified after Fig. 1 inTack et al. (1994). The Archaean craton shown is a part of the Tanzanian craton; the Palaeoproterozoic is called the Buganda-Toro Belt in Uganda; the Kibaran granites are peraluminous (the representation includes pre-Kibaran Palaeoproterozoic relicts and post-Kibaran Sn-bearing granites); the Kabanga–Musongati (KM) mafic and ultramafic alignment and the A-type granitoids intrude the ‘‘Boundary Zone’’ between the Western Internal Domain (WID) and the Eastern External Domain (EED); the Neoproterozoic comprises the Bukoban–Malagarasian sedimentary and volcanic basin. K: Kabanga; M: Musongati; Mr: Muremera; Ny: Nyabikere; M-B: Mukanda-Buhoro; W: Waga; R: Rutovu; N-S: Nyange-Songa and Mg: Mugina. Inset: C = Congo; U = Uganda; R = Rwanda; B = Bur-undi; T = Tanzania.
upwards into alternating pyroxenites and peridotites), which pass into a mafic zone dominated by noritic rocks. Further to the west, a septum of metasediments clearly separates a norite unit belonging to the Muson-gati body from a gabbronoritic unit which is the south-ern extension of the Mukanda-Buhoro body (Fig. 2).
The rocks show typical characteristics of cumulates (Fig. 3). Olivine is euhedral in ultramafic rocks and the first liquidus mineral to appear in the sequence of rocks (Fig. 4). Fine-grained chromite is present in minor amounts in ultramafic rocks. In the dunite and lherzolite units, it occurs in the interstices between cumulus oli-vine, and never forms chromitite layers. This points to an intercumulus status. Plagioclase laths and prismatic orthopyroxene crystals define a conspicuous lamination in noritic and gabbroic rocks. In ultramafic rocks plag-ioclase shows an intercumulus habit. Sulphides (pyrrho-tite, pentlandite, chalcopyrite) are disseminated in all
rocks as minor components, but can locally attain a sig-nificant content in the dunite and lherzolite units. Fe–Ti
Fig. 2. Schematic geological map of the Mukanda-Buhoro and Musongati bodies. Inset: alignment of the two types of layered bodies in Burundi (the Muremera body, north of the area, is not shown).
Fig. 3. Cumulate textures: (A) Peridotite (MU75) showing euhedral olivine (partly serpentinised); plagioclase, pyroxenes and opaques (chromite and sulphides) are intercumulus minerals; uncrossed nicols. (B) Gabbronorite (BU21) with clear igneous lamination, underlined by plagioclase laths; interstitial Ca-poor pyroxene (inverted pigeonite) forms large oikocrysts; uncrossed nicols. Lengths of the photos are 1.3 cm.
Fig. 4. Stratigraphic succession of petrographic units and mineral chemistry variation (microprobe data of Deblond, 1994) in the Mukanda-Buhoro–Musongati layered intrusion. Cumulus minerals are represented by a continuous line, intercumulus minerals by a dashed line. Question marks indicate ambiguous cumulus/intercumu-lus status. Intercumucumulus/intercumu-lus Ca-poor pyroxene in the lherzolite and dunite units is usually altered. In brackets, number of determinations. Note that the Mwiriba quartz norite unit and the Mutanga amphibole norite unit, defined byDeblond (1994), are located in the upper part of the norite unit (Fig. 2) and have essentially the same cumulus mineral compositions.
oxide minerals are present in the two upper units, and form a deposit in the Mukanda area.
It emerges fromFigs. 2 and 4that rocks more evolved than anorthosite are not known in the Mukanda-Bu-horo intrusion. They might have been removed by ero-sion or faulting. The most evolved rock in all KM intrusions of Burundi has been found in the Mugina intrusion. It is an anorthosite (An56) with fayalitic oli-vine (Fo7) and inverted pigeonite (mg# 17) (Deblond, 1994).
It is noteworthy that gabbronorites in all considered KM intrusions are texturally similar to the Bushveld Main Zone cumulates, described by Wager and Brown (Fig. 213, p. 387 inWager and Brown, 1968): slightly re-versely zoned tabular plagioclases with local myrmekite define an igneous lamination. Inverted pigeonite also forms large poikiloblasts (up to 10 cm in diameter), as is also the case in the Bushveld Complexvon Gru¨new-aldt (1990).
Cryptic layering is conspicuous in cumulus mineral compositions (Fig. 4). Cumulus olivine varies from Fo90 to Fo76 and cumulus plagioclase from An90 to An76. This evolution is parallel to the mg# variation of both pyroxenes. These characteristics strongly suggest that fractional crystallization was the major differentia-tion mechanism operating in the various magma cham-bers, as also noted by Evans et al. (1999) for the Kabanga intrusion.
4. Cumulate chemistry, the chromitite issue, and sulphide segregation
Whole rock chemical compositions of rocks coming from the Mukanda-Buhoro and Musongati (MBM) contiguous bodies, and from the other massifs, have been reported byDeblond (1994). Representative analy-ses of the main rock types are given in Table 1. Fig. 5
illustrates the evolution with all available data, for a few selected elements.
As expected the whole rock compositions are strongly controlled by the nature and proportion of the cumulus minerals. For instance, Al contents (Fig. 5) are very low in dunite. In lherzolite and pyroxenite it mainly repre-sents the small amount of intercumulus plagioclase. In norite and gabbronorite, where plagioclase is a cumulus mineral, the Al2O3contents range between 10% and 20% to reach a maximum of 27% in anorthosite.
The Cr-evolution is somewhat unexpected (Fig. 5) be-cause Cr is typically a compatible element in basaltic magma, which usually sharply decreases in the succes-sive liquids and cumulates of a fractional crystallization process. Here Cr tends to increase with declining MgO in the olivine-bearing ultramafic rocks. Cr concentra-tions culminate in the pyroxenite and rapidly decrease in norite to disappear in anorthosite. The most
impor-tant Cr-carrier minerals in ultramafic rocks are chromite
and pyroxenes. Deblond (1994) has shown that the
Cr2O3 contents of Ca-rich and Ca-poor pyroxenes can be as high as 1.0% (sample RU12.53) and 0.73% (sample MU69), respectively. Considering Cr mineral/melt parti-tion coefficient values of 34 for clinopyroxene and 10 for orthopyroxene (Rollinson, 1993), it can be inferred that the Cr melt composition was of the order of several hun-dred ppm when the pyroxenes appeared at the liquidus. These high values are reflected in the high Cr content of pyroxenite inFig. 5. Actually, the behaviour of Cr in du-nite and lherzolite shows a grossly defined increase, which is partly due to the increase with differentiation of the amount of Cr-rich pyroxene in the cumulate, and partly to the increase of the Cr content in the melt. Whatever the influence of the two factors, this behav-iour certainly points to a melt not depleted in Cr. This in turn puts strong constraints on the value of the Cr bulk partition coefficient between dunite and lherzolite cumulates and melt. This must thus have evolved at val-ues close to one, or even lower than one, which drasti-cally limits the fraction of chromite in the cumulate. In other words, if chromite was an abundant liquidus min-eral in the dunite and lherzolite cumulates, it would have determined a very high bulk partition coefficient for Cr and a rapid depletion of Cr in the successive melts. Chr-omite is indeed only found sporadically in dunite and lherzolite and always as a trace mineral. Though our data do not preclude occurrence of chromitite associated with pyroxenite, no chromitite layers analogous to those found in the Bushveld Complex or the Great Dyke have ever been observed. The present considerations suggest that there is a low probability to find chromitite depos-its, thus limiting the Cr metallogenic potential of the studied layered bodies in Burundi.
The Ni-content shows an overall decrease with MgO (Fig. 5).Fig. 6shows that there is no Ni–S correlation, thus indicating that the sulphides are not systematically Ni-rich. It also shows that Ni might be contained by pentlandite in some samples. Moreover, the high Ni val-ues of two samples with very low S contents nevertheless indicate that, in these samples at least, olivine is Ni-rich. More data on olivine composition are obviously needed, but the present work suggests that the whole rock Ni-content might be controlled by olivine. The high Ni val-ues observed in dunite (3000–4000 ppm) fall within the range commonly found in layered intrusions (seeSimkin and Smith, 1970). This suggests that an immiscible Ni-rich sulphide melt did not segregate from the magma prior to its intrusion into the chamber. Such segregation would have depleted the coexisting silicate melt in chalc-ophile elements and PGE, and olivine would have been strongly depleted in Ni (Thompson and Naldrett, 1984; Naldrett, 1989). Therefore, unlike the occurrence of the Kabanga Ni-sulphide deposits in Tanzania (Evans et al., 1999, 2000), large bodies of massive Ni–Cu–Co sulphide
Table 1
Whole-rock major (%) and trace (ppm) element analyses of selected petrographic types from Kabanga–Musongati intrusions
Dunite Dunite Peridotite Peridotite Pyroxenite Pyroxenite Norite Norite Norite Gabbronorite Gabbronorite Anorthosite Anorthosite
F211.54* F211.53* MU74* MU79A* MU69 NY.P128 MU51 MU39 MU61 BU30 BU105** MU31 BU85
SiO2 38.34 37.69 44.10 43.12 53.19 51.33 52.57 49.28 49.18 50.89 49.65 47.56 51.98 Ti02 0.03 0.01 0.06 0.04 0.15 0.18 0.16 0.05 1.31 0.28 0.29 0.18 0.06 Al2O3 0.30 0.38 1.45 4.33 3.36 6.16 8.22 20.96 16.76 15.21 1549 2623 25.71 Fe2O3 13.44 13.15 14.24 15.37 11.86 13.59 13.52 6.82 13.66 13.34 12.65 5.91 3.33 MnO 0.16 0.17 0.21 0.17 0.18 0.20 0.23 0.10 0.24 0.20 0.19 0.10 0.04 MgO 47.30 47.74 38.88 32.43 28 05 24.27 20.32 10.27 5.60 7.49 7.03 3.86 0.99 CaO 0.22 0.10 0.04 3.96 2.36 4.88 5.45 11.74 10.62 10.86 11.87 14.55 13.21 Na2O 0.10 0.01 0.07 0.01 001 0.25 0.37 0.85 1.46 2.08 1.97 1.33 3.49 K2O 0.01 0.01 0.01 0.01 0.01 0.05 0.03 0.15 0.57 0.08 0.21 0.18 0.33 P2O5 0.01 0.01 0.01 0.01 0.01 0.01 0.02 0.04 0.05 0.01 0.02 0.02 0.01 Total 99.90 99.27 99.07 99.44 99.14 100.93 100.89 100.26 99.46 100.43 99.36 99.92 99.16 S (%) 1.23 0.0779 0.003 0.47 0.039 0.28 0.08 0.26 0.03 0.27 0.04 0.11 Zr 7 3 3 7 10 3 3 13 62 3 4.5 12 18 Hf 0.5 0.2 0.2 0.2 0.2 1.6 0.2 024 0.2 0.2 Nb 3 7 3 3 3 3 3 3 9 3 0.44 3 3 Ta 0.2 0.5 0.5 0.5 0.5 0.5 0.5 0.06 0.5 0.5 Rb 3 3 3 3 3 3 3 6 29 3 3.4 9 10 Sr 3 3 3 21 12 47 49 113 162 160 148 179 267 Ba 15 15 41 41 15 15 43 15 174 34 56 47 81 Ni 4048 2983 3377 1591 132 52 12 12 12 12 61 12 12 V 28 12 82 94 414 225 455 208 373 406 337 156 77 Cr 1704 2790 2746 3604 5027 3576 1710 1176 77 104 152 199 12 Zn 53 44 62 70 74 86 78 36 76 68 66 35 27 Co 155 164 150 143 100 92 40 12 37 65 43 12 12 La 0.2 1.0 0.9 1 1 3.1 13.1 1.3 1.9 3.2 3.3 Ce 4 4 5 1.5 6 27 15 3.0 6 7 Nd 2 02 2 2.5 2 12 2 2.2 2 2 Sm 0.1 0.2 0.2 0.2 0.5 2.5 0.5 0.79 0.4 0.6 Eu 0.1 0.2 0.2 0.5 0.7 1.0 0.5 0.67 1.0 0.8 Gd 1.02 Tb 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.19 0.2 0.2 Yb 0.1 0.4 0.5 0.2 0.5 1.7 0.7 0.97 0.3 0.4 Lu 0.02 0.09 0.09 0.1 0.07 0.27 0.12 0.15 0.09 0.07 [La/Yb]n 1.4 1.8 1.3 3.9 4.4 5.5 1.3 1.4 7.6 5.9 Eu/Eu* 0.8 1.3 1.3 3.2 3.0 1.4 2.2 2.2 4.9 3.1
Major elements by XRF analysis (CIGI, ULG) on calcinated samples; trace element at X-ray Assay Lab., Canada, except samples *by XRF and **by ICP MS (CIGI. ULG); S by XRF at X-ray Assay Lab., Canada.
Sample location: F211.54: Musongati body, bore hole 211, depth 54 m; F211.53: same, depth 53 m; MU74: Musongati body, Nyamabuye river;MU79A: Musongati body, Marumanga; MU69: Musongati body 1.5 km W of Nkeyuke; NY.P128: Nyabikere body, Ntaruka River, NE of Nyarurnazi; MU51:Musongati body, Miremera; MU39: Musongati body, Muhoza; MU61: Musongati body, Gihehe river 1km W of Miremera; BU30: Buhoro body, 500 m SW of Nyarurambi ; BU105: Buhoro body, Gihamagara Mu31: Musongati body; 500 m W of Nyarubimba; BU85: Buhoro body, Mutukura River, 1 km E of Shungwe.
J.-C. Duchesne et al. / Journal of African Earth Sciences 39 (2004) 133–145 137
are not to be expected in the Musongati and Waga bod-ies in Burundi. On the other hand, S-saturation started during fractionation of dunite and gave rise to dissemi-nated sulphides, which are depleted in Ni.
5. Strontium initial ratio in cumulate rocks
Sr isotopes have been studied in the various intrusions in order to identify different magma influxes (Kruger, 1994) and variations due to crustal assimila-tion. Detailed data are reported in and summarized in
Fig. 7. The interval of variation of the Sr initial ratio, all massifs included, is relatively restricted: a first group contains 17 samples out of 21 and ranges from 0.7068 to 0.7089, and a second group of four samples ranges be-tween 0.7104 and 0.7120 (Fig. 7). There is no definite relationship with the intrusion or with the rock type,
Fig. 5. Variation in whole-rock compositions: Al2O3(%), Cr (ppm) and Ni (ppm) vs. MgO (%) (data fromDeblond, 1994).
Fig. 6. S (%) vs. Ni (ppm) in ultramafic rocks (data fromDeblond, 1994). Same symbols as inFig. 5. The dashed line represents the Ni/S ratio in pentlandite (Fe,S)9S8with Fe/Ni = 1, the richest Ni sulphide carrier in these rocks. Samples below this line have a Ni content not completely accounted for by pentlandite. Conversely, samples above the line contain sulphides other than pentlandite, i.e. pyrrhotite, pyrite and chalcopyrite. Samples on the line may contain pentlandite as the only Ni carrier. Note that six other samples with S > 1.4% (see inset) also fall in the S-rich field. Samples F276.53 and MU74 (Table 1) are very poor in S, thus Ni must be essentially contained in olivine.
Fig. 7. Histogram of whole rock Sr isotope initial ratios at 1275 Ma (data fromDeblond, 1993, 1994).
as illustrated in Fig. 8. The highest initial ratios are found in some noritic rocks, and the most evolved cumulates, i.e. anorthosites, are not different from the least evolved gabbronorite and pyroxenite. These char-acteristics suggest (1) that the high Sr initial ratios
(sec-ond group) probably result from strong local
contamination with country rocks, as also observed by
Evans et al. (1999) in the Kabanga intrusion, and (2) that continuous assimilation during fractional crystalli-zation (AFC process) was not significant. Variations within bodies are not related to structural height in the intrusion and distinct influxes of magma with different isotopic signature cannot be identified. The high average Sr initial ratio in group 1 thus appears to be related to the characteristics of the magma source rather than to an AFC process.
6. Geochemistry of related fine-grained rocks
No chilled margins have been identified in the various intrusions. The identification of possible melts, either parental or more evolved, has thus to rely on indirect evidence. Fine-grained rocks, with typical subophitic texture or with microgranular texture occur within the intrusions, preferentially close to the margins. In two
cases, these fine-grained rocks belong to dykes cutting across the country rocks. Though the ages of these dykes are unknown, their occurrence at small distance from the main intrusions (Rutovu, RU04; S. Mukanda-Bu-horo, 93BU4 and 93BU5) and their textural resem-blance to the facies within the bodies strongly suggest that they belong to the same magmatic event. Classically such fine-grained rocks offer the best possible opportu-nity to define melts belonging to the same kindred as those forming the layered bodies. More evolved rocks, characterized by intersertal granophyric structure
(A117B and NYP176; Table 2), have also been
considered.
Major and trace element geochemistry as well as iso-topes are used to check the cumulus vs. melt status of some of these rocks and the nature of the magma series to which they belong, in order to test possible contami-nation processes during differentiation, and to define the nature of the magma source. Data on major and trace elements are reported inTable 2and isotopes inTable 3. 6.1. Major and trace elements
Out of the 16 cases considered (Table 2), the fine-grained microgabbros (SG23C, BU92, BU78 and RU10) cannot be considered as possible melts. They dis-play a distinct positive Eu anomaly (Fig. 9) and low incompatible element contents. They show similarities with cumulate rocks and probably represent crystal– laden liquids. The two granophyric rocks (A117B and NYP176) are typically more differentiated with higher SiO2, K2O, Rb, Ba, HFSE and REE contents as well as negative Eu anomalies (Fig. 9andTable 2). The rest of the samples can be divided into three groups, geo-graphically and geochemically distinct. The first group comprises four fine-grained rocks related to the Muk-anda-Buhoro body. This group is relatively enriched in silica (48–53% SiO2) and LREE, and depleted in TiO2, Sr and Nb (Fig. 10A and B). The mg# is close to 60. The second group is made up of fine-grained rocks from the Nyabikere area (Fig. 10C and D). Com-pared to the Mukanda-Buhoro group, the rocks are ri-cher in Ti and Nb (no anomalies in spidergrams), the REE distribution has a constant slope, and the silica content (48%) and the mg# (50) are lower. The third group includes fine-grained rocks from Waga and Rut-ovu. The major elements are comparable to the Nyabik-ere group, but they show spidergrams depleted in incompatible elements and horizontal REE distribu-tions (Fig. 10C and D).
6.2. Isotopes
Sr and Nd isotopic compositions have been measured on several melts. Table 3 shows the new data together with previously published data for some cumulate rocks
Fig. 8. Sr isotope initial ratio vs. MgO (%) whole-rock content. (A) Rock types, same symbols as inFig. 5; (B) intrusions.
Table 2
Whole-rock major (%) and trace (ppm) element analyses of cumulate and finegrained rocks from Kabanga–Musongati intrusions
Type SG23C BU92 BU78 RU10 BU93 93BU4 93BU5 BU17 A117B NY P176 NY P142 NY P145 NY P138 NY80 W26 RU04
cum cum cum cum Buhoro Buhoro Buhoro Buhoro Rutovu Nyabikere Nyabikere Nyabikere Nyabikere Nyabikere Waga Rutovu
SiO2 49.88 49.08 49.70 51.03 52.44 53.01 52.88 47.81 55.20 61.32 48.53 48.59 47.72 48.18 47.91 50.69 TiO2 0.61 0.41 027 0.58 0.44 0.57 0.65 0.76 0.49 1.63 1.72 1.63 1.71 1.61 1.20 1.44 Al2O3 14.63 15.53 1560 14.59 15.95 13.75 1455 15.62 14.84 13.83 14.14 14.15 12.93 14.00 13.49 13.77 Fe2O3 14.15 13.86 11.71 14.78 9.50 10.49 10.35 12.21 10.34 12.46 13.96 13.09 14.21 14.09 14.10 13.02 MnO 0.20 0.22 0.19 0.20 0.14 0.16 0.21 0.20 0.15 0.18 0.23 0.20 0.23 0.26 0.23 0.19 MgO 7.89 8.18 7.87 7.00 7.16 8.85 7.49 10.42 6.30 1.92 6.80 7.43 7.48 6.57 7.96 7.12 CaO 12.54 11.93 13.50 10.43 11.05 9.40 9.30 11.05 9.20 5.80 11.83 12.28 12.06 12.43 12.42 11.72 Na2O 1.54 1.33 1.35 2.02 1.33 0.90 1.13 1.07 1.95 1.78 2.59 2.38 2.34 1.57 1.82 1.53 K2O 0.13 0.08 0.16 0.19 0.22 0.27 0.46 0.16 1.29 1.20 0.34 0.32 0.29 0.31 0.13 0.19 P2O5 0.01 0.02 0.01 0.02 0.21 0.14 0.27 0.07 0.15 0.36 0.20 0.18 0.18 0.13 0.10 0.12 Total 101.58 100.64 100.36 100.84 98.44 97.54 97.29 99.37 99.91 100.48 100.34 100.25 99.15 99.15 99.36 99.79 U 0.08 0.15 0.12 0.13 0.7 1.2 1.3 1.1 3.1 0.8 0.5 0.4 0.4 0.2 0.2 Th 0.06 0.43 0.25 0.39 1.5 3.6 3.7 0.9 4.3 11.8 1.7 1.0 1.5 1.4 0.4 Zr 8 11 12 32 66 78 82 79 106 232 105 103 113 106 65 73 Hf 0.32 0.42 0.55 1.0 2.3 2.6 2.9 2.0 3.5 6.6 2.9 3.2 3.0 3.1 1.6 2.3 Nb 0.08 1.5 0.7 3.2 2.8 3.7 4.1 6.7 3.0 25 17 17 21 19 < 4 4.7 Ta 0.02 0.06 0.14 0.24 0.35 0.44 0.47 0.40 0.50 1.73 1.00 < 1 1.32 1.26 < 1 0.34 Rb 3.7 1.7 4.7 4.0 14 25 23 8 51 72 12 11 11 9 2.5 2.1 Sr 137 130 163 139 118 98 126 117 216 160 203 222 223 216 126 128 Ba 35 55 61 101 130 90 150 99 331 277 144 128 128 135 30 66 Ni 27 9 12 31 12 60 12 68 113 93 113 100 57 V 413 259 128 220 221 245 230 272 149 55 299 267 287 254 311 286 Cr 386 206 247 184 265 604 384 630 51 47 120 290 205 219 180 83 Zn 77 82 77 103 55 68 53 75 84 120 79 72 108 117 82 71 Co 44 36 26 41 29 39 36 46 12 10 55 55 51 47 57 35 Cu 17 18 18 503 19 95 152 35 Ga 16 12 10 17 23 19 19 16 Y 7.7 9.5 15.0 15.0 19.0 22.9 23.5 22.0 30.0 57.4 34.0 31.0 27.9 26.2 30.0 24.2 La 1.82 3.84 3.80 7.71 11.50 13.00 13.80 11.40 20.57 59.14 14.50 15.00 12.49 12.07 4.10 4.64 Ce 2.94 6.30 7.59 15.66 23.80 27.90 29.60 25.86 43.11 88.87 31.00 33.00 28.39 27.55 11.00 11.87 Pr 0.47 0.84 1.23 1.88 3.40 4.00 4.20 3.32 5.51 14.07 3.77 3.70 1.99 Nd 2.28 3.78 5.80 8.82 12.80 14.40 14.90 13.72 22.47 59.82 16.00 17.00 17.17 16.63 8.00 9.73 Sm 0.78 1.06 1.84 2.28 2.60 3.10 3.30 3.14 4.50 13.25 3.60 3.70 4.69 4.31 2.40 3.13 Eu 0.62 0.87 0.79 0.96 0.74 0.95 0.96 0.91 1.17 3.49 1.90 1.50 1.73 1.64 0.90 1.20 Gd 107 1.34 2.36 2.39 2.50 3.40 3.40 3.47 4.63 12.52 5.11 4.80 3.86 Tb 0.20 0.25 0.41 0.40 0.40 0.64 0.71 0.50 1.88 0.80 0.70 0.83 0.82 0.80 0.65 Dy 1.29 1.54 2.69 2.73 2.70 3.20 3.50 3.66 4.71 10.59 4.97 4.78 4.31 Ho 0.30 0.35 0.60 0.61 0.68 0.82 0.82 0.73 0.90 2.03 1.06 0.98 0.89 Er 0.86 1.08 1.76 1.76 1.90 2.10 2.32 2.15 2.86 5.64 2.94 2.73 2.64 Tm 0.12 0.18 0.26 0.29 0.29 0.34 0.78 0.45 0.39 0.42 140 J.-C. Duchesne et al. / Journal of African Earth Sciences 39 (2004) 133–145
(Tack et al., 1994). Sr and Nd data, calculated for an age of 1275 Ma, are plotted in Fig. 11. In the case of a 1370 Ma age for the emplacement of the KM intrusions, eNdvalues would be modified by less thanjej, whereas Sr initial ratios would not be significantly modified because of low Rb contents. These variations would thus not af-fect our conclusions. Two groups of samples distinctly appear: the first one, with negative eNd(1275 Ma) values and high Sr initial compositions, includes the Muk-anda-Buhoro melts, defined above, and cumulates from the Mukanda-Buhoro and Musongati bodies. The sec-ond group, with depleted mantle values, comprises the Nyabikere and the Waga fine-grained rocks (the Rutovu dolerite with high Sr isotope ratio and positive eNdhas been perturbed and should be ignored).
Interestingly, the cumulate samples from Nyabikere (Table 3) show high Sr initial ratios, similar to those in the other layered intrusions. It can be concluded that the Mukanda-Buhoro melts and cumulates are comag-matic, and, on the other hand, that the Nyabikere melts are not related to the cumulates of this intrusion. As for the Waga intrusions, isotopic data on the cumulate rocks are lacking, thus precluding a comparison with the doleritic rock.
These data corroborate previous conclusions regard-ing the existence of two distinct differentiation trends
and sources for the Kabanga-Musongati layered
intrusions and for the A-type granitoids of Burundi, respectively (Tack et al., 1994). We confirm that, paradoxically, the Kabanga–Musongati source is an enriched one, possibly the subcontinental lithospheric mantle, and that the A-type granitoid source is a depleted asthenospheric mantle. The depleted source of the Nyabikere and Waga dolerites indicates that they are not linked to the Kabanga–Musongati ma-fic–ultramafic magmatism. Their source characteristics could suggest that they are related to the A-type grani-toid magmatism, but this link has to be further investigated.
6.3. Modelling the melt–cumulate relationships
Sample BU17 from the Mukanda-Buhoro group has already been considered as a possible ‘‘chilled’’ magma (Tack et al., 1994). Since this sample, with 10.4% MgO, is the most primitive composition of the Muk-anda-Buhoro group that we have identified, it is worth-while determining the nature and composition of its liquidus minerals. Modelling of this composition using MIXNFRAC algorithm (Nielsen, 1989) and assuming a 3 kbar pressure and oxygen fugacity at the FMQ buf-fer, is reported in Table 4. It can be seen that the first mineral to appear at the liquidus is Fo88, a value match-ing the olivine composition in lherzolite (seeFig. 4). This liquid can thus produce a lherzolite cumulate similar to those of Musongati. It should be noted, however, that
V b 0.90 1.13 1.85 1.81 1.84 1.98 2.21 2.17 2.96 4.99 2.70 2.60 2.81 2.71 2.10 2.69 Lu 0.14 0.17 0.26 0.27 0.34 0.33 0.36 0.32 0.45 0.71 0.39 0.39 0.40 0.40 0.31 0.37 Eu/E u* 2.08 2.23 1.16 1.25 0.88 0.89 0.87 0.84 0.78 0.82 1.43 1.15 1.08 1.10 0.88 1.06 mg# 53 54 57 49 60 63 59 63 55 24 49 53 51 48 53 52 M ajor ele ment by XRF on calcinated sam ples and tr ace eleme nts by IC P–MS (CIGI, ULG ). Sa mple loca tion and type: SG23C: Nya nga-So nga intru sion, roa d Nya nge-Mo sso, boulde rs of microgranul ar gabbro ; BU92: Buhoro intru sion, Giham agara valle y, bould ers of mic rogran ular gab bron orite; BU78: Buhoro intrusio n, E o f Gihamba valleys, bould ers of mesoc ratic fin e-graine d mas sive gabbro norite; RU10: Ruto vu intrusio n, Gas enyi valle y, boulde rs of mic rogran ular gab bron orites (with pyr oxene oik ocrysts); BU 93: Buhoro massif , valle y S of Giham araga, microgab bron orite bo ulders ; 93BU4 and 93B U5: S M ukan da-Bu ho ro massif, 2 k m S E o f Mwishanga do lerite dikes cro sscuting the Nyama buye sub-formatio n; BU 17: Buhoro massif , M pame, bridge over rive r Nyakijanda, suboph itic mic rogabbro boulde rs; A117 B: Ruto vu massif , brid ge over Gas enyi rive r, amph ibole bearin g dolerite; NY P176: Nya bikere intru sion, NE Rwanda garo, granoph yre; NY P1 42: Nyabike re intrusio n, Ntar uka rive r, subo phitic microgab bro boulde rs; NY P145: Nya bikere intru sion, Ntar uka rive r, subo phitic microgab bro boulde r; NY P138: Nyabik ere intru sion, Ntar uka river, dolerite boulde r; NY80: Nya bik ere intru sion, N o f Nyaru nazi, afflu ent to Ntar uka st ream. Olivine bearin g subo phitic microgabbro; W 26: Waga intrusio n, 1 k m Wof Kabu ndugu, subo phitic microgabbro fragmen ts in laterite; RU 04: Rutovu massif , sourc es of the Nile, do lerile dyke .
Table 3
Sr and Nd isotope data from fine-grained rocks and cumulates from Kabanga–Musongati intrusions Rb (ppm) Sr (ppm) 87Rb/86Sr 87Sr/86Sr ± 2rm(*106) ISr(1275 Ma) Sm (ppm) Nd (ppm) 147Sm/144Nd 143Nd/144Nd ± 2rm(*106) eNd(1275 Ma) TCHUR TDM Fine-grained rocks Mukanda-Buhoro group BU17* 8.3 116.6 0.2061 0.712980 ± 30 0.709215 15.70 64.10 0.14812 0.511987 ± 9 4.78 2307 2468 BU93 14.4 118.0 0.3535 0.718671 ± 10 0.712212 2.60 12.80 0.12283 0.511776 ± 14 4.76 2043 2126 93 BU5 23.3 126.0 0.5359 0.721624 ± 9 0.711834 3.30 14.90 0.13393 0.511868 ± 13 4.78 2141 2249 93 BU4 25.1 98.2 0.7408 0.723933 ± 9 0.710398 3.10 14.40 0.13018 0.511886 ± 15 3.82 2025 2112 Nyabikere group NY 80 9.5 215.6 0.1273 0.704109 ± 10 0.701783 4.31 16.63 0.15685 0.512565 ± 10 5.10 1274 NYP142 12.0 203.0 0.1710 0.704594 ± 10 0.701470 3.60 16.00 0.13608 0.512545 ± 10 8.11 1002 982 NYP145 11.0 222.0 0.1433 0.704435 ± 8 0.701816 3.70 17.00 0.13164 0.512541 ± 9 8.76 960 938
Waga and Rutovu group
W 26 2.5 126.0 0.0574 0.703076 ± 8 0.702027 2.40 8.00 0.18146 0.512837 ± 9 6.39 1026 RU 04 2.1 127.6 0.0484 0.713017 ± 11 0.712132 3.13 9.73 0.19491 0.512835 ± 11 4.15 1544 Cumulates BU30* 1.2 157.2 0.0221 0.708900 ± 50 0.708496 0.73 2.24 0.19709 0.512233 ± 29 7.99 5291 BU78* 4.7 162.9 0.0835 0.7118S0 ± 30 0.710354 2.38 8.00 0.17992 0.512193 ± 20 5.96 3296 MU39* 5.4 112.8 0.1386 0.713580 ± 30 0.711048 0.67 3.13 0,12944 0.511759 ± 10 6.18 2216 2329 NYP128 2.10 47.3 0.1253 0.711080 ± 50 0.708788 NYP141 6.70 134 0.1447 0.711060 ± 40 0.708416 NY33 5.90 71.2 0.2386 0.712122 ± 13 0.707762 NY65 11.40 205 0.1611 0.713363 ± 9 0.710419 NYP141-Plag 6.60 255 0.0747 0.709964 ± 8 0.708600
Sample location and rock type: BU30: Buhoro intrusion. SW of Nyarurambi, layered gabbronorite cumulate; BU78: Buhoro intrusion, SSW of Gihamba, fine-grained gabbronorite; MU39; Musongati intrusion, Nyamabuye stream near Muhoza, noritic cumulate boulders; NYP128: Nyabikere intrusion. Ntaruka river NNE of Nyarunazi, pyroxenite (boulders); NYP141: Nyabikere intrusion, norite cumulate (boulders): NYP141: Plag; Nyabikere intrusion, plagioclase separated from the previous samples; NY33: Nyabikere intrusion, S of Rwakagari, fine-grained gabbronorite (boulders); NY65: Nyabikere intrusion. Ntaruka river S of Mubanga, gabbronorite (boulders); Rock type and location of all other samples: seeTable 2. TDMmodel ages are given for samples having147Sm/144Nd < 0.15.
Samples marked by an *are fromTack et al. (1994).
142 J.-C. Duchesne et al. / Journal of African Earth Sciences 39 (2004) 133–145
olivine crystallised alone only over a short interval (4%), which precludes the formation of large quantities of ultramafic cumulates. Moreover, the composition of the olivine in dunite is slightly more Mg rich (Fo90), which calls for a melt richer in MgO than BU17. It can thus be concluded that BU17 is not parental to the Musongati body, but is already more evolved. This in turn suggests that the parental magma tends towards a picritic composition.
Other compositions (BU93 and NYP) have also been modelled and the results are reported inTable 4. Plagio-clase is the first liquidus mineral in both cases and there-fore these melts are not suitable as parent liquids for the ultramafic intrusions nor for the gabbronorite intrusions (the calculated plagioclase An78for NYP is too anorth-ite-rich).
It should be further noted that the three melts which have been modelled produce troctolite over large inter-vals of crystallization, and not pyroxenite and norite. This feature is actually an artefact of NielsenÕs algo-rithm: the model artificially delays the appearance of liq-uidus orthopyroxene, as observed in the Bushveld Complex (Cawthorn, pers. comm.) and in several other cases (Duchesne, unpubl.).
Fig. 10. Chondrite-normalized trace element composition of fine-grained rocks: (A, B) Mukanda-Buhoro fine-grained rocks; (C,D) Nyabikere, Rutovu and Waga fine-grained rocks. Spidergram normalizing values fromThompson et al. (1984).
Fig. 9. Chondrite-normalized REE composition in granophyric rocks (filled square) and in fine-grained microgabbros (open circle).
7. The two magma hypothesis
Though it is difficult to assess the composition of the parental magmas, due to the lack of chilled margin and adequate dyke rocks, some general features point to the possibility of two distinct magma compositions having been involved in the evolution of the KM mafic and ultramafic intrusions.
The marked difference of bulk compositions of the intrusions is notable and not due to a possible variation of the level of erosion. The intrusions dip60° to the west so that both the floor and roof of the magma cham-bers can be observed in some places despite locally developed thrust faulting in the contact zones. As already pointed out, some bodies (e.g. Mugina, Nyange-Songa, Rutovu, Mukanda-Buhoro) comprise only anorthosite and gabbronorite, with the latter lithol-ogy lying directly on the floor (Fig. 2). The Waga body consist exclusively of ultramafic rocks. In Muremera, Nyabikere and Musongati, ultramafic rocks dominate
over plagioclase-bearing rocks, and, in the latter body, dunite, lherzolite and pyroxenite-peridotite represent more than 66% of the area (Deblond, 1994).
The parental magma of the Musongati body was likely to have been picritic, as also suggested for the Ka-banga intrusion (Evans et al., 2000), thus reducing the availability of evolved residual liquids necessary to form gabbronoritic and anorthositic cumulates. Nowhere has a transition from norite to gabbronorite been observed in the mafic intrusions. Indeed, the gabbronorites of the Mukanda-Buhoro body are clearly separated by a septum (Fig. 2) of metasediments from the most evolved norites of the Musongati body. In other words, it ap-pears that the geographical contiguity between the Muk-anda-Buhoro and the Musongati bodies is coincidental rather than petrogenetically significant. A relatively large gap in mineral composition is observed (Fig. 4) be-tween the Musongati norites (An89-86and opx mg# 79– 75 in norite) and the Mukanda-Buhoro gabbronorites (An75-58and opx mg# 58–45).
None of the previous characteristics provide decisive evidence for the existence of two parental magmas. But if this hypothesis is accepted, the analogies with the Bushveld Complex, already mentioned for the petro-graphical similarities between gabbronorites from both occurrences, become even stronger. Indeed, several magma influxes have been invoked to account for the huge volume of the Bushveld Complex (see discussion in Cawthorn, 1996). Particularly two influxes seem to be important there: the Lower and Critical Zone mag-ma, probably of picritic composition, and the Main Zone magma, more evolved in composition, and of huge lateral extension. We incline to the view that similar magmas can be invoked in the KM alignment in Bur-undi. The first magma would be a picritic basalt, but, contrary to the Bushveld Complex, it was unable to pro-duce chromitite (e.g. Musongati, Waga). The second one, giving rise to gabbronorite and anorthosite, would be analogous to the Main Zone magma (e.g. Mugina, Nyange-Songa, Rutovu, Mukanda-Buhoro) and also shows a large geographical extension. In contrast to the Bushveld case, there is, however, not much difference in isotopic signature between the two magmas. Appar-ently, they came from isotopically similar sources. Alternatively it is not precluded that the second magma, parental to the mafic intrusions, might be derived from the picritic magma by fractional crystallisation in a deeper level magma chamber. If this is the case, the absence of gabbronoritic rocks and anorthosites in the ultramafic intrusions nevertheless remains intriguing.
8. Conclusions
The Kabanga–Musongati layered intrusions are dis-tributed into ultramafic and gabbronoritic types of
bod-Table 4
MIXNFRAC modelling of fractional crystallization (3 kbar and fO2at FMQ) of possible parental magmas
BU17 (Mukanda-Buhoro type, Mg-rich) 0% 1257°C Fo88(peridotite)
4% 1228°C Fo85+ An86(troctolite)
46% 1162°C Fo75+ An83+ Aug (olivine gabbro) BU93 (Mukanda-Buhoro type, Mg-poor) 0% 1233°C An84(anorthosite)
6% 1200°C An82+ Fo85(troctolite)
28% 1167°C An80+ Fo81+ Aug (olivine gabbro) NYP (average of 4 samples) (Nyabikere type) 0% 1204°C An78(anorthosite)
5% 1188°C An77+ Fo80(troctolite)
20%1160°C An74+ Fo76+ Aug (olivine gabbro)
Fig. 11. eNd(1275 Ma)vs. Sr initial ratio (1275 Ma) in Mukanda-Buhoro, Nyabikere, Waga and Rutovu fine-grained rocks and in Mukanda-Buhoro and Musongati cumulate rocks. Values from Bukirasazi mafic rocks (Tack et al., 1994) are also given for comparison.
ies, which both form an alignment in a weakness zone between the Kibaran belt and the Tanzania craton (Tack et al., 1994, 2002). The intrusions display, when considered as a whole, a classical series of cumulates from dunite to anorthosite with well-defined cryptic lay-ering (Deblond, 1994). Fractional crystallization was the dominant differentiation mechanism. The Cr–Ni behav-iour in the melt precludes the occurrence of large chro-mitite and Ni sulphide concentrations associated with the ultramafic rocks.
Several lines of evidence point to the occurrence of two types of magma to form the various bodies. The first one, a picritic basalt, has given rise to the ultramafic bodies. The second one, analogous to the Main Zone magma of the Bushveld Complex, has formed the gab-bronorite bodies. Both magmas came from the same en-riched mantle source, characterized by negative eNd values and high Sr isotope initial ratios, indicative of an old subcontinental lithospheric mantle.
In the same weakness zone, A-type granitoids and doleritic dykes have also been emplaced but they possess a clearly distinct depleted asthenospheric mantle signature.
Acknowledgments
Major and trace element analyses were performed at the CIGI (University of Lie`ge) by G. Bologne. The Bel-gian ‘‘Direction Ge´ne´rale pour la Coope´ration au De´veloppement’’ (DGCD) is thanked for supporting the Doctorat thesis of A. Deblond (BDI/CI 14497/11) and a University Cooperation Programme with the Uni-versity of Burundi at Bujumbura. The paper has bene-fited from constructive reviews of J. R. Wilson and D. M. Evans, who are sincerely thanked.
References
Cawthorn, R.G., 1996. Layered Igneous Rocks. Elsevier, Amsterdam, 531p.
Deblond, A., 1993. Ge´ologie et pe´trologie des massifs basiques de la ceinture Kabanga-Musongati au Burundi. Doctorat dissertation. University of Lie`ge, Belgium, 235p.
Deblond, A., 1994. Ge´ologie et pe´trologie des massifs basiques et ultrabasiques de la ceinture Kabanga-Musongati au Burundi. Annales du Muse´e Royal de lÕAfrique Centrale Tervuren Sciences ge´ologiques 99, 1–123.
Deblond, A., Tack, L., 1999. Main characteristics and review of mineral resources of the Kabanga–Musongati alignment in Bur-undi. Journal of African Earth Sciences 29, 313–328.
Evans, A.M., 1993. Ore Geology and Industrial Minerals. An Introduction. Blackwell Science, Oxford, 389p.
Evans, D.M., Byemelwa, L., Gilligan, J., 1999. Variability of magmatic sulphide compositions at the Kabanga nickel prospect, Tanzania. Journal of African Earth Sciences 29, 329–351.
Evans, D.M., Boadi, I., Byemelwa, J., Gilligan, J., Kabete, J., Marcet, P., 2000. Kabanga magmatic nickel suphide deposits, Tanzania: morphology and geochemistry of associated intrusions. Journal of African Earth Sciences 30, 651–674.
Kruger, F.J., 1994. The Sr-isotopic stratigraphy of the west-ern Bushveld Complex. South African Journal of Geology 97, 393–398.
Lee, C.A., 1996. A review of mineralization in the Bushveld Complex and some other layered mafic intrusions. In: Cawthorn, R.G. (Ed.), Layered Intrusions. Elsevier, Amsterdam.
Naldrett, A.J., 1989. Magmatic Sulfide Deposits. Oxford University Press, Oxford, 186p.
Naldrett, A.J., Cameron, G., von Grunewaldt, G., Sharpe, M.R., 1987. The formation of stratiform PGE deposits in layered intrusions. In: Parsons, I. (Ed.), Origin of Igneous Layering, NATO ASI Series. Reidel, Dordrecht, pp. 313–398.
Nielsen, R.L., 1989. Phase equilibria constraints on liquid lines of descent generated by paired assimilation and fractional crystalli-zation: trace elements and Sr and Nd isotopes. Journal of Geophysical Research 94, 787–794.
Rollinson, H., 1993. Using Geochemical Data: Evaluation, Presenta-tion, Interpretation. Longman Scientific and Technical, Harlow, 352p.
Simkin, T., Smith, J.V., 1970. Minor element distribution in olivine. Journal of Geology 78, 304–325.
Tack, L., Deblond, A., 1990. Intrusive character of the Late Kibaran magmatism in Burundi. IGCP No. 255 Newsletter 3, 81–87. Tack, L., Liegeois, J.-P., Deblond, A., Duchesne, J.C., 1994. Kibaran
A-type granitoids and mafic rocks generated by two mantle sources in a late orogenic setting (Burundi). Precambrian Research 68, 323– 356.
Tack, L., Fernandez-Alonso, M., Tahon, A., Wingate, M., Barritt, S., 2002. The ‘‘Northeastern Kibaran Belt’’ (NKB) and its mineral-isations reconsidered: new constraints from a revised lithostratig-raphy, a GIS-compilation of existing geological maps and a review of recently published as well as unpublished igneous emplacement ages in Burundi. 11th Quadrennial IAGOD Symposium and GEOCONGRESS 2002, 22–26 July 2002, Windhoek, Namibia, Extended Abstract on CD-Rom and Synopsis in Conference Programme, p. 42.
Thompson, J.F.H., Naldrett, A.J., 1984. Sulfide–silicate reactions as a guide to Ni–Cu–Co mineralization in central Maine. In: Buchanan, D.L., Jones, M.J. (Eds.), Sulfide Deposits in Mafic and Ultramafic Rocks. Institute of Mining and Metallurgy Special Publication, London, pp. 103–113.
Thompson, R.N., Morrison, M.A., Hendry, G.L., Parry, S.J., 1984. An assessment of the relative roles of crust and mantle in magma genesis: an elemental approach. Philosophical Transactions of the Royal Society of London A310, 549–590.
von Gru¨newaldt, G., 1990. On the phase change orthopyroxene pigeonite and the resulting textures in the Main and Upper zones of the Bushveld Complex in Eastern Transvaal. Geological Society of South Africa Special publication 1, 67–73.
Wager, L.R., Brown, G.M., 1968. Layered igneous rocks. Oliver and Boyd Ltd, London, 588p.
